A high quality speech synthesis method based on interpolating the voiced/unvoiced (V/UV) mixture functions [1] of the multiband excitation model (MBE) [2,3] is proposed. In MBE model, each harmonic band of fundamental frequency in an excitation spectrum is rigidly declared as either voiced or unvoiced while it should be a mixture of two. In the proposed method, each harmonic band in a short time spectrum is synthesized by mixing both voiced and unvoiced energies. The ratio of the V/UV energy in a spectrum is determined by the V/UV mixture function which is subsequently parametized by an all-zero model. Smooth transition of excitation between phonetic units can be achieved by linearly interpolating the V/UV mixture functions of adjacent frames.